Swiss Startup Flexion Robotics Develops AI System Enabling Humanoid Robots to Perform Complex Office Tasks

Swiss startup Flexion Robotics, founded by former Nvidia researchers, has developed an advanced artificial intelligence system that enables humanoid robots to autonomously perform complex office tasks. Demonstrated recently, their robot can follow natural language instructions to collect packages from a mailroom, navigate stairs, operate elevators, unpack contents, and store items in designated drawers without human intervention. This capability is powered by a combination of imitation learning and reinforcement learning, where robots first learn basic skills in simulation by observing humans and then refine their performance through trial and error. A central AI model breaks down complex commands into actionable sequences, with CEO Nikita Rodin highlighting reinforcement learning as the "secret ingredient" that elevates the system's performance beyond competitors relying on remote control or fixed scripts.

Flexion showcased its technology at the ICRA 2026 conference, reporting a success rate exceeding 95% across 300 autonomous trials. The company recently secured $50 million in funding led by international investors and is expanding operations into the United States. Rather than producing specific robots, Flexion aims to become a software "brain" provider, offering a unified platform compatible with various humanoid robots. This approach could accelerate the integration of robots into the workforce regardless of hardware manufacturer.

This breakthrough comes amid a surge of over $10 billion invested in humanoid robotics in the past year alone, intensifying the question of when robots will become true colleagues rather than mere tools in the workplace.